On Sun, 24 May 2020 00:30:21 -0600, Jeff Barnett <jbb@notatt.com>
wrote:

> ...
> By the way, these "use" directives are about as low
>level as you can get. Why doesn't the system load whatever is used? Is
>your toy designed for less then megabyte memory machines. [There's two
>questions buried in here.]

Stupid question: why do you consider Scheme using a library to be more
"low level" than Lisp using a package?

Gauche is a bit odd in that it has both "use" and "require", but the
difference is only syntax. Gauche has a culture of distinguishing use
of the built-in extensions ("use") vs other libraries ("require").

On 2024-07-08, George Neuner <gneuner2@comcast.net> wrote:
> On Sun, 24 May 2020 00:30:21 -0600, Jeff Barnett <jbb@notatt.com>
> wrote:
>
>> ...
>> By the way, these "use" directives are about as low
>>level as you can get. Why doesn't the system load whatever is used? Is
>>your toy designed for less then megabyte memory machines. [There's two
>>questions buried in here.]
>
> Stupid question: why do you consider Scheme using a library to be more
> "low level" than Lisp using a package?
>
> Gauche is a bit odd in that it has both "use" and "require", but the
> difference is only syntax. Gauche has a culture of distinguishing use
> of the built-in extensions ("use") vs other libraries ("require").

The built-in stuff should not require any magic incantation to come
online. If such a thing is needed in something calling itself a Lisp
dialect, that's embarrassing.

--
TXR Programming Language: http://nongnu.org/txr
Cygnal: Cygwin Native Application Library: http://kylheku.com/cygnal
Mastodon: @Kazinator@mstdn.ca

On Tue, 9 Jul 2024 02:18:23 -0000 (UTC), Kaz Kylheku
<643-408-1753@kylheku.com> wrote:

>On 2024-07-08, George Neuner <gneuner2@comcast.net> wrote:
>> On Sun, 24 May 2020 00:30:21 -0600, Jeff Barnett <jbb@notatt.com>
>> wrote:
>>
>>> ...
>>> By the way, these "use" directives are about as low
>>>level as you can get. Why doesn't the system load whatever is used? Is
>>>your toy designed for less then megabyte memory machines. [There's two
>>>questions buried in here.]
>>
>> Stupid question: why do you consider Scheme using a library to be more
>> "low level" than Lisp using a package?
>>
>> Gauche is a bit odd in that it has both "use" and "require", but the
>> difference is only syntax. Gauche has a culture of distinguishing use
>> of the built-in extensions ("use") vs other libraries ("require").
>
>The built-in stuff should not require any magic incantation to come
>online. If such a thing is needed in something calling itself a Lisp
>dialect, that's embarrassing.

Clearly Scheme is a Lisp *derivative*, but I think it's debatable
whether Scheme really can be considered a Lisp *dialect*. Its
semantics are very different and it was as much inspired by Algol as
by Lisp.

YMMV.

In any case, in that code fragment, the functions provided by
util.match are *not* built-in - they are in an implementation specific
library.

Prior to R6RS (circa 2007), Scheme did not even have syntax for
specifying or using libraries. Although a number of implementations
did provide them, there was no "standard" way to do it. Gauche's
"extensions" fall into this category.

George Neuner <gneuner2@comcast.net> writes:
> Prior to R6RS (circa 2007), Scheme did not even have syntax for
> specifying or using libraries. Although a number of implementations
> did provide them, there was no "standard" way to do it.

That is an intentional practice with Scheme, I believe. They don't like
to standardize stuff until there are implementations out there that
users seem to think are done right.

On Thu, 11 Jul 2024 01:45:39 -0700, Paul Rubin
<no.email@nospam.invalid> wrote:

>George Neuner <gneuner2@comcast.net> writes:
>> Prior to R6RS (circa 2007), Scheme did not even have syntax for
>> specifying or using libraries. Although a number of implementations
>> did provide them, there was no "standard" way to do it.
>
>That is an intentional practice with Scheme, I believe. They don't like
>to standardize stuff until there are implementations out there that
>users seem to think are done right.

R_RS documents are not standards per se - they simply describe what is
considered a conforming implementation.

There are ISO and ANSI standards for Scheme, but they are based on
R3RS, circa ~1990[*]. As far as I'm aware, no other version of scheme
has been the subject of a standard.

[*] The R3RS document is not dated (that I can find), but I was using
an R3RS implementation in grad school in the early 90's.

On Fri, 12 Jul 2024 07:19:47 -0400, George Neuner
<gneuner2@comcast.net> wrote:

>[*] The R3RS document is not dated (that I can find), but I was using
>an R3RS implementation in grad school in the early 90's.

I think by the time I was leaving grad school in '93 they were up to
R5RS already.

George Neuner <gneuner2@comcast.net> wrote:
> On Fri, 12 Jul 2024 07:19:47 -0400, George Neuner
> <gneuner2@comcast.net> wrote:
>
>>[*] The R3RS document is not dated (that I can find), but I was using
>>an R3RS implementation in grad school in the early 90's.
>
> I think by the time I was leaving grad school in '93 they were up to
> R5RS already.

Interesting, the R3RS indeed has no date. The RRRS is dated 1985 and
the R4RS 1991. In think R3RS may have been 1986 or 1987. FWIW, I see
no date later than 1986 in the references. Then there was a long break
between R4 and R5. R5RS appeared in 1998.

--
Nils M Holm < n m h @ t 3 x . o r g > http://t3x.org

On 12 Jul 2024 15:37:33 GMT, Nils M Holm <nmh@sraddha.invalid> wrote:

>George Neuner <gneuner2@comcast.net> wrote:
>> On Fri, 12 Jul 2024 07:19:47 -0400, George Neuner
>> <gneuner2@comcast.net> wrote:
>>
>>>[*] The R3RS document is not dated (that I can find), but I was using
>>>an R3RS implementation in grad school in the early 90's.
>>
>> I think by the time I was leaving grad school in '93 they were up to
>> R5RS already.
>
>Interesting, the R3RS indeed has no date. The RRRS is dated 1985 and
>the R4RS 1991. In think R3RS may have been 1986 or 1987. FWIW, I see
>no date later than 1986 in the references. Then there was a long break
>between R4 and R5. R5RS appeared in 1998.

I defer to your knowledge of the history. 8-)

I thought it was an R5 I was using because it had syntax-rules, but
re-reading the R_RS docs, I see that R4 could have had them as an
extension.

George Neuner <gneuner2@comcast.net> wrote:
> I defer to your knowledge of the history. 8-)

8)

> I thought it was an R5 I was using because it had syntax-rules, but
> re-reading the R_RS docs, I see that R4 could have had them as an
> extension.

R4RS described define-syntax, syntax-rules, etc in an appendix. I think
the system was pretty much identical to what became "official" in R5RS,
but I am not entirely sure, because I have always prefered the low-level,
LISP defmacro-style macros, which many R4RS implementations also offered.

--
Nils M Holm < n m h @ t 3 x . o r g > http://t3x.org